Efficient Compression Techniques For 3D Meshes Using Wavelet Transform

In order to render 3D objects more realistically, 3D mesh model produces a dramatic amount of data, which comprises of a big problem for storage and web transmission. A substantial compression is hence essential for the data of 3D objects to reduce the storage, save network bandwidth and shorten transmission delay. Besides, an efficient compression can also strengthen the interactive capability of 3D graphics.Taking account of the strong correlations among the 3D mesh vertex coordinates and the concentration of the symbols referencing the connectivity, an efficient compression scheme for 3D meshes using wavelet transform was then proposed to improve the compression bit-rate of 3D meshes with the help of energy concentration, decorrelation and multi-resolution relation.The first algorithm proposed here is the single rate compression based on the wavelet transform, which applies an MAPS algorithm on the irregular original mesh to obtain a semi-regular mesh in a process of remeshing. Then wavelet transform will be used to execute a multi-resolution decomposition generating an irregular coarse mesh (base mesh) and a sequence of wavelet coefficients with the hierarchical geometry details. An improved Edgebreaker algorithm and arithmetic coding are separately employed to the coding of the base mesh and the quantization of wavelet coefficients. Both the encoding results will finally be written into one unified file with a designed format. By tested with the lifting Loop wavelet, Butterfly wavelet and the lifting Butterfly wavelet, the compression ratio of this approach shows a better performance than Edgebreaker algorithm. However the time cost for the compression and decompression is larger due to the remeshing and wavelet transform.A novel algorithm of progressive compression based on wavelet transform proposed in this paper deals with the connectivity of the original mesh and consequently reconstructs the basic topological relations of the remeshed base mesh. Not only the spatial correlations between the different levels of details but also the time correlations are exploited to enhance the compression ratio. A concept of"frame", which represents different levels of detail at different time, is introduced into the static mesh. A compression mode of F+3D (a 3D mesh within a temporal frame) is then achieved. The temporal correlations among the frames at different time will be decreased by using a temporal-based lifting wavelet. Then, geometry details in a frame will be translated into several wavelet coefficients which will be joined into the bit stream after quantization and encoding.Experimental results suggest that the algorithm achieve a good compression ratio for larger 3D models, and that the performance be satisfactory in the process of progressive decoding.